Andrey-Tkachev / content-aware-resize

Real time seam carving algorithm implementation

About

This is an implementation of the real time content aware resize algorithm. Project consist from console and gui applications (cxinar and uxinar respectively) and library xinar for C++, which provide resize services.

Dependencies and building

Xinar builds as shared library so you don't need anything to build it from sources. But you have to have opencv2 on your PC and OpenCV_LIBS linked to your application, wich uses xinar.

To build console or graphic application from source you need

• Qt5 core (only for uxinar)
• Qt5 widgets (only for uxinar)
• libboost-program-options (only for cxinar)
• libboost-filesystem (only for cxinar)
• libboost-system (only for cxinar)
• OpenCV 2

Project uses CMake>=3.5 as building system.

To build project from source in your building folder type:

:~$ cmake <path to content-aware-resize sources>
:~$ make xinar
:~$ make uxinar
:~$ make cxinar

Usage

You can use cxinar in two ways: as a C++ library and as an application.

Application usage

Once you built cxinar, you can use it in 2 ways: as a console application or via GUI.

Console usage example:

Resize image examle.jpg to 600 by 600 pixels:

:~$ ./cxinar -i example.jpg -w 600 -h 600

All commands are available via -? (--help) flag.

GUI usage example:

Just run uxinar as executable and choose file to resize in the file menu. To resize change size of the window.

NOTE: Xinar can resize picture up to 2 times in both height and width while expanding picture.

Library usage

xinar::Resize is main class that provides resize services. Let input and output be cv::Mat matrix objects, size be cv::Size object. Then you can apply content aware resize in following way:

...
#include "xinar.h"
...

int main() {
    ...

    // For single usage
    xinar::resize(cv_image, cv_output, cv_size);
    
    // For multiple usage
    xinar::Resize resizer(cv_image);
    for (auto size : cv_sizes) {
        resizer.process(cv_output, size);
        ...
    }

    ...
}

All documentation is available in docs folder.

Structure

• core -- seam carving algorithm implementation
• cui -- console application code
• filters -- filters functors implementation
• gui -- GUI application code
• include -- internal headers
• interface -- external headers
• utils -- various code to make life happier

Copyright Andery Tkachev, Nikita Orlov, Dmitriy Malygin, 2017.